cannam@167: /* cannam@167: * Copyright (c) 2003, 2007-14 Matteo Frigo cannam@167: * Copyright (c) 2003, 2007-14 Massachusetts Institute of Technology cannam@167: * cannam@167: * This program is free software; you can redistribute it and/or modify cannam@167: * it under the terms of the GNU General Public License as published by cannam@167: * the Free Software Foundation; either version 2 of the License, or cannam@167: * (at your option) any later version. cannam@167: * cannam@167: * This program is distributed in the hope that it will be useful, cannam@167: * but WITHOUT ANY WARRANTY; without even the implied warranty of cannam@167: * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the cannam@167: * GNU General Public License for more details. cannam@167: * cannam@167: * You should have received a copy of the GNU General Public License cannam@167: * along with this program; if not, write to the Free Software cannam@167: * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA cannam@167: * cannam@167: */ cannam@167: cannam@167: /* This file was automatically generated --- DO NOT EDIT */ cannam@167: /* Generated on Thu May 24 08:06:26 EDT 2018 */ cannam@167: cannam@167: #include "rdft/codelet-rdft.h" cannam@167: cannam@167: #if defined(ARCH_PREFERS_FMA) || defined(ISA_EXTENSION_PREFERS_FMA) cannam@167: cannam@167: /* Generated by: ../../../genfft/gen_r2cf.native -fma -compact -variables 4 -pipeline-latency 4 -n 12 -name r2cf_12 -include rdft/scalar/r2cf.h */ cannam@167: cannam@167: /* cannam@167: * This function contains 38 FP additions, 10 FP multiplications, cannam@167: * (or, 30 additions, 2 multiplications, 8 fused multiply/add), cannam@167: * 21 stack variables, 2 constants, and 24 memory accesses cannam@167: */ cannam@167: #include "rdft/scalar/r2cf.h" cannam@167: cannam@167: static void r2cf_12(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) cannam@167: { cannam@167: DK(KP866025403, +0.866025403784438646763723170752936183471402627); cannam@167: DK(KP500000000, +0.500000000000000000000000000000000000000000000); cannam@167: { cannam@167: INT i; cannam@167: for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) { cannam@167: E T5, Tp, Tm, Tk, Ty, Tt, Ta, Tq, Tn, Tf, Tz, Tu, Tl, To; cannam@167: { cannam@167: E T1, T2, T3, T4; cannam@167: T1 = R0[0]; cannam@167: T2 = R0[WS(rs, 2)]; cannam@167: T3 = R0[WS(rs, 4)]; cannam@167: T4 = T2 + T3; cannam@167: T5 = T1 + T4; cannam@167: Tp = FNMS(KP500000000, T4, T1); cannam@167: Tm = T3 - T2; cannam@167: } cannam@167: { cannam@167: E Tg, Th, Ti, Tj; cannam@167: Tg = R1[WS(rs, 1)]; cannam@167: Th = R1[WS(rs, 3)]; cannam@167: Ti = R1[WS(rs, 5)]; cannam@167: Tj = Th + Ti; cannam@167: Tk = FNMS(KP500000000, Tj, Tg); cannam@167: Ty = Ti - Th; cannam@167: Tt = Tg + Tj; cannam@167: } cannam@167: { cannam@167: E T6, T7, T8, T9; cannam@167: T6 = R0[WS(rs, 3)]; cannam@167: T7 = R0[WS(rs, 5)]; cannam@167: T8 = R0[WS(rs, 1)]; cannam@167: T9 = T7 + T8; cannam@167: Ta = T6 + T9; cannam@167: Tq = FNMS(KP500000000, T9, T6); cannam@167: Tn = T8 - T7; cannam@167: } cannam@167: { cannam@167: E Tb, Tc, Td, Te; cannam@167: Tb = R1[WS(rs, 4)]; cannam@167: Tc = R1[0]; cannam@167: Td = R1[WS(rs, 2)]; cannam@167: Te = Tc + Td; cannam@167: Tf = FNMS(KP500000000, Te, Tb); cannam@167: Tz = Td - Tc; cannam@167: Tu = Tb + Te; cannam@167: } cannam@167: Cr[WS(csr, 3)] = T5 - Ta; cannam@167: Ci[WS(csi, 3)] = Tt - Tu; cannam@167: Tl = Tf - Tk; cannam@167: To = Tm - Tn; cannam@167: Ci[WS(csi, 1)] = FMA(KP866025403, To, Tl); cannam@167: Ci[WS(csi, 5)] = FNMS(KP866025403, To, Tl); cannam@167: { cannam@167: E Tx, TA, Tv, Tw; cannam@167: Tx = Tp - Tq; cannam@167: TA = Ty - Tz; cannam@167: Cr[WS(csr, 5)] = FNMS(KP866025403, TA, Tx); cannam@167: Cr[WS(csr, 1)] = FMA(KP866025403, TA, Tx); cannam@167: Tv = T5 + Ta; cannam@167: Tw = Tt + Tu; cannam@167: Cr[WS(csr, 6)] = Tv - Tw; cannam@167: Cr[0] = Tv + Tw; cannam@167: } cannam@167: { cannam@167: E Tr, Ts, TB, TC; cannam@167: Tr = Tp + Tq; cannam@167: Ts = Tk + Tf; cannam@167: Cr[WS(csr, 2)] = Tr - Ts; cannam@167: Cr[WS(csr, 4)] = Tr + Ts; cannam@167: TB = Ty + Tz; cannam@167: TC = Tm + Tn; cannam@167: Ci[WS(csi, 2)] = KP866025403 * (TB - TC); cannam@167: Ci[WS(csi, 4)] = KP866025403 * (TC + TB); cannam@167: } cannam@167: } cannam@167: } cannam@167: } cannam@167: cannam@167: static const kr2c_desc desc = { 12, "r2cf_12", {30, 2, 8, 0}, &GENUS }; cannam@167: cannam@167: void X(codelet_r2cf_12) (planner *p) { cannam@167: X(kr2c_register) (p, r2cf_12, &desc); cannam@167: } cannam@167: cannam@167: #else cannam@167: cannam@167: /* Generated by: ../../../genfft/gen_r2cf.native -compact -variables 4 -pipeline-latency 4 -n 12 -name r2cf_12 -include rdft/scalar/r2cf.h */ cannam@167: cannam@167: /* cannam@167: * This function contains 38 FP additions, 8 FP multiplications, cannam@167: * (or, 34 additions, 4 multiplications, 4 fused multiply/add), cannam@167: * 21 stack variables, 2 constants, and 24 memory accesses cannam@167: */ cannam@167: #include "rdft/scalar/r2cf.h" cannam@167: cannam@167: static void r2cf_12(R *R0, R *R1, R *Cr, R *Ci, stride rs, stride csr, stride csi, INT v, INT ivs, INT ovs) cannam@167: { cannam@167: DK(KP866025403, +0.866025403784438646763723170752936183471402627); cannam@167: DK(KP500000000, +0.500000000000000000000000000000000000000000000); cannam@167: { cannam@167: INT i; cannam@167: for (i = v; i > 0; i = i - 1, R0 = R0 + ivs, R1 = R1 + ivs, Cr = Cr + ovs, Ci = Ci + ovs, MAKE_VOLATILE_STRIDE(48, rs), MAKE_VOLATILE_STRIDE(48, csr), MAKE_VOLATILE_STRIDE(48, csi)) { cannam@167: E T5, Tp, Tb, Tn, Ty, Tt, Ta, Tq, Tc, Ti, Tz, Tu, Td, To; cannam@167: { cannam@167: E T1, T2, T3, T4; cannam@167: T1 = R0[0]; cannam@167: T2 = R0[WS(rs, 2)]; cannam@167: T3 = R0[WS(rs, 4)]; cannam@167: T4 = T2 + T3; cannam@167: T5 = T1 + T4; cannam@167: Tp = FNMS(KP500000000, T4, T1); cannam@167: Tb = T3 - T2; cannam@167: } cannam@167: { cannam@167: E Tj, Tk, Tl, Tm; cannam@167: Tj = R1[WS(rs, 1)]; cannam@167: Tk = R1[WS(rs, 3)]; cannam@167: Tl = R1[WS(rs, 5)]; cannam@167: Tm = Tk + Tl; cannam@167: Tn = FNMS(KP500000000, Tm, Tj); cannam@167: Ty = Tl - Tk; cannam@167: Tt = Tj + Tm; cannam@167: } cannam@167: { cannam@167: E T6, T7, T8, T9; cannam@167: T6 = R0[WS(rs, 3)]; cannam@167: T7 = R0[WS(rs, 5)]; cannam@167: T8 = R0[WS(rs, 1)]; cannam@167: T9 = T7 + T8; cannam@167: Ta = T6 + T9; cannam@167: Tq = FNMS(KP500000000, T9, T6); cannam@167: Tc = T8 - T7; cannam@167: } cannam@167: { cannam@167: E Te, Tf, Tg, Th; cannam@167: Te = R1[WS(rs, 4)]; cannam@167: Tf = R1[0]; cannam@167: Tg = R1[WS(rs, 2)]; cannam@167: Th = Tf + Tg; cannam@167: Ti = FNMS(KP500000000, Th, Te); cannam@167: Tz = Tg - Tf; cannam@167: Tu = Te + Th; cannam@167: } cannam@167: Cr[WS(csr, 3)] = T5 - Ta; cannam@167: Ci[WS(csi, 3)] = Tt - Tu; cannam@167: Td = KP866025403 * (Tb - Tc); cannam@167: To = Ti - Tn; cannam@167: Ci[WS(csi, 1)] = Td + To; cannam@167: Ci[WS(csi, 5)] = To - Td; cannam@167: { cannam@167: E Tx, TA, Tv, Tw; cannam@167: Tx = Tp - Tq; cannam@167: TA = KP866025403 * (Ty - Tz); cannam@167: Cr[WS(csr, 5)] = Tx - TA; cannam@167: Cr[WS(csr, 1)] = Tx + TA; cannam@167: Tv = T5 + Ta; cannam@167: Tw = Tt + Tu; cannam@167: Cr[WS(csr, 6)] = Tv - Tw; cannam@167: Cr[0] = Tv + Tw; cannam@167: } cannam@167: { cannam@167: E Tr, Ts, TB, TC; cannam@167: Tr = Tp + Tq; cannam@167: Ts = Tn + Ti; cannam@167: Cr[WS(csr, 2)] = Tr - Ts; cannam@167: Cr[WS(csr, 4)] = Tr + Ts; cannam@167: TB = Ty + Tz; cannam@167: TC = Tb + Tc; cannam@167: Ci[WS(csi, 2)] = KP866025403 * (TB - TC); cannam@167: Ci[WS(csi, 4)] = KP866025403 * (TC + TB); cannam@167: } cannam@167: } cannam@167: } cannam@167: } cannam@167: cannam@167: static const kr2c_desc desc = { 12, "r2cf_12", {34, 4, 4, 0}, &GENUS }; cannam@167: cannam@167: void X(codelet_r2cf_12) (planner *p) { cannam@167: X(kr2c_register) (p, r2cf_12, &desc); cannam@167: } cannam@167: cannam@167: #endif